Prep data

Load necessary packages

setwd("~/Desktop/working-with-lyle/Formality_Project")#set our WD 
if (!require("pacman")) install.packages("pacman") #run this if you don't have pacman 
library(pacman)
pacman::p_load(tidyverse,rlang, zoo, lubridate, plotrix, ggpubr, caret, broom, kableExtra, reactable, install = T) 
#use pacman to load packages quickly

Define Aesthetics for graphs and stuff

palette_map = c("#3B9AB2", "#EBCC2A", "#F21A00")
palette_condition = c("#ee9b00", "#bb3e03", "#005f73")

plot_aes = theme_classic() +
  theme(legend.position = "top",
        legend.text = element_text(size = 12),
        text = element_text(size = 16, family = "Futura Medium"),
        axis.text = element_text(color = "black"),
        axis.line = element_line(colour = "black"),
        axis.ticks.y = element_blank())

Define Table Functions

 table_model = function(model_data) {
   model_data %>% 
     tidy() %>% 
     rename("SE" = std.error,
            "t" = statistic,
            "p" = p.value) %>%
     kable() %>% 
     kableExtra::kable_styling()
 }

Load data

df <- read_csv('Atlantic_Cleaned_all_vars.csv') #read in the data

Tidy the data

 tidy_df <- df %>%
   group_by(Date) %>% ###grouping by the year 
  mutate_at(vars("Analytic","WPS","BigWords","Period","readability","grade_level"), as.numeric) %>% 
   summarise_at(vars("Analytic","WPS","BigWords","Period","readability","grade_level"),  funs(mean, std.error),) 

# Get the mean values for the year 1932
year_means <- tidy_df %>%
  filter(Date == 1857) 

#create centered variables on 1857
tidy_df$Analytic_centered <- tidy_df$Analytic_mean - 85.94
tidy_df$WPS_centered <- tidy_df$WPS_mean - 37.14
tidy_df$BigWords_centered <- tidy_df$BigWords_mean - 25.68
tidy_df$Period_centered <- tidy_df$Period_mean - 4.589
tidy_df$readability_centered <- tidy_df$readability_mean - 57.45
tidy_df$grade_level_centered <- tidy_df$grade_level_mean - 10.71

Corpus Summary Stats

Dates

df %>% 
  select(Date) %>% 
  range()
## [1] 1857 2022

Raw count of Articles

df %>%
  select(Filename) %>%
  dplyr::summarize(n = n()) %>%
  reactable::reactable(striped = TRUE)

Part of Speech Graphs

Plot the Smoothed Data

Plotting each individual book along with the line of best fit (loess regression) to see trends in the data.

#Plot our smoothed data 

#we are using Non-tidy data here to capture the individual variation 

#Analytic Thinking 

Analytic_smooth <- ggplot(data=df, aes(x=Date, y=Analytic, group=1)) +
  ggtitle("Analytic Thinking") +
  geom_point(color = "dodgerblue3", alpha = 0.15) + 
  geom_smooth(method = "loess", span = 0.50 )+ 
  plot_aes +
  labs(x = "Year", y = '% of Total Words') +
  theme(axis.text.x=element_text(angle=45, hjust=1), 
        plot.title.position = 'plot', 
        plot.title = element_text(hjust = 0.5, face = "bold", size = 16)) +
  theme(plot.title.position = 'plot', 
        plot.title = element_text(hjust = 0.5, face = "bold", size = 16)) + 
  theme(axis.text=element_text(size=16),
        axis.title=element_text(size=20,face="bold"))+
  theme(plot.title.position = 'plot', 
        plot.title = element_text(hjust = 0.5, face = "bold", size = 20)) +
  theme(axis.text=element_text(size = 14),
        axis.title=element_text(size = 20,face="bold")) +
  annotate(geom="text",x=1860,
             y=25,label="
             estimate = 0.1
             p-value <. 001
           
           ", size = 3.5)

#Bigwords
Bw_smooth <- ggplot(data=df, aes(x=Date, y=BigWords, group=1)) +
  ggtitle("Big Words (Letters > 6)") +
  geom_point(color = "dodgerblue3", alpha = 0.15) + 
  geom_smooth(method = "loess", span = 0.50 )+ 
  plot_aes +
  labs(x = "Year", y = '% of Total Words') +
  theme(axis.text.x=element_text(angle=45, hjust=1), 
        plot.title.position = 'plot', 
        plot.title = element_text(hjust = 0.5, face = "bold", size = 16)) +
  theme(plot.title.position = 'plot', 
        plot.title = element_text(hjust = 0.5, face = "bold", size = 16)) + 
  theme(axis.text=element_text(size=16),
        axis.title=element_text(size=20,face="bold"))+
  theme(plot.title.position = 'plot', 
        plot.title = element_text(hjust = 0.5, face = "bold", size = 20)) +
  theme(axis.text=element_text(size = 14),
        axis.title=element_text(size = 20,face="bold")) +
  annotate(geom="text",x=1860,
             y=40,label="
             estimate = 0.0701
             p-value < .001
           
           ", size = 3.5)

#Periods
period_smooth <- ggplot(data=df, aes(x=Date, y=Period, group=1)) +
  ggtitle("Period Usage") +
  geom_point(color = "dodgerblue3", alpha = 0.15) + 
  geom_smooth(method = "loess", span = 0.50 )+ 
  plot_aes +
  labs(x = "Year", y = '% of Total Words') +
  theme(axis.text.x=element_text(angle=45, hjust=1), 
        plot.title.position = 'plot', 
        plot.title = element_text(hjust = 0.5, face = "bold", size = 16)) +
  theme(plot.title.position = 'plot', 
        plot.title = element_text(hjust = 0.5, face = "bold", size = 16)) + 
  theme(axis.text=element_text(size=16),
        axis.title=element_text(size=20,face="bold"))+
  theme(plot.title.position = 'plot', 
        plot.title = element_text(hjust = 0.5, face = "bold", size = 20)) +
  theme(axis.text=element_text(size = 14),
        axis.title=element_text(size = 20,face="bold")) +
  annotate(geom="text",x=1860,
             y=15,label="
             estimate = 0.0204
             p-value < .001
           
           ", size = 3.5)

#words per sentence
wps_smooth <- ggplot(data=df, aes(x=Date, y=WPS, group=1)) +
  ggtitle("Words per Sentence") +
  geom_point(color = "dodgerblue3", alpha = 0.15) + 
  geom_smooth(method = "loess", span = 0.50 )+ 
  plot_aes +
  labs(x = "Year", y = '# of Words') +
  theme(axis.text.x=element_text(angle=45, hjust=1), 
        plot.title.position = 'plot', 
        plot.title = element_text(hjust = 0.5, face = "bold", size = 16)) +
  theme(plot.title.position = 'plot', 
        plot.title = element_text(hjust = 0.5, face = "bold", size = 16)) + 
  theme(axis.text=element_text(size=16),
        axis.title=element_text(size=20,face="bold"))+
  theme(plot.title.position = 'plot', 
        plot.title = element_text(hjust = 0.5, face = "bold", size = 20)) +
  theme(axis.text=element_text(size = 14),
        axis.title=element_text(size = 20,face="bold")) +
  annotate(geom="text",x=1860,
             y=200,label="
             estimate = -0.0497
             p-value < .001
           
           ", size = 3.5)


smooth_graphs <- ggpubr::ggarrange(Analytic_smooth,Bw_smooth,period_smooth,wps_smooth,
                                   ncol=2, nrow=2, common.legend = TRUE, legend = "bottom")
annotate_figure(smooth_graphs,
                top = text_grob("Smooth Formality Graphs",  color = "black", face = "bold", size = 20),
                bottom = text_grob(
                "Note. Horizontal shading represents Standard Error."
                                   , color = "Black",
                                   hjust = 1.05, x = 1, face = "italic", size = 14))

Plotting the Smoothed Data (by year)

Analytic_smooth_tidy <- ggplot(data=tidy_df, aes(x=Date, y=Analytic_mean, group=1)) +
  ggtitle("Analytic Thinking") +
  geom_point(color = "dodgerblue3", alpha = 0.7) + 
  geom_smooth(method = "loess", span = 0.50 )+ 
  plot_aes +
  labs(x = "Year", y = '% of Total Words') +
  theme(axis.text.x=element_text(angle=45, hjust=1), 
        plot.title.position = 'plot', 
        plot.title = element_text(hjust = 0.5, face = "bold", size = 16)) +
  theme(plot.title.position = 'plot', 
        plot.title = element_text(hjust = 0.5, face = "bold", size = 16)) + 
  theme(axis.text=element_text(size=16),
        axis.title=element_text(size=20,face="bold"))+
  theme(plot.title.position = 'plot', 
        plot.title = element_text(hjust = 0.5, face = "bold", size = 20)) +
  theme(axis.text=element_text(size = 14),
        axis.title=element_text(size = 20,face="bold")) +
  annotate(geom="text",x=1860,
             y=95,label="
             estimate = 0.0782
             p-value < .001
           
           ", size = 3.5)

#Bigwords
Bw_smooth_tidy <- ggplot(data=tidy_df, aes(x=Date, y=BigWords_mean, group=1)) +
  ggtitle("Six letter > N words") +
  geom_point(color = "dodgerblue3", alpha = 0.7) + 
  geom_smooth(method = "loess", span = 0.90 )+ 
  plot_aes +
  labs(x = "Year", y = '% of Total Words') +
  theme(axis.text.x=element_text(angle=45, hjust=1), 
        plot.title.position = 'plot', 
        plot.title = element_text(hjust = 0.5, face = "bold", size = 16)) +
  theme(plot.title.position = 'plot', 
        plot.title = element_text(hjust = 0.5, face = "bold", size = 16)) + 
  theme(axis.text=element_text(size=16),
        axis.title=element_text(size=20,face="bold"))+
  theme(plot.title.position = 'plot', 
        plot.title = element_text(hjust = 0.5, face = "bold", size = 20)) +
  theme(axis.text=element_text(size = 14),
        axis.title=element_text(size = 20,face="bold")) +
  annotate(geom="text",x=1860,
             y=30,label="
             estimate = 0.0431
             p-value < .001
           
           ", size = 3.5)

#Periods
period_smooth_tidy <- ggplot(data=tidy_df, aes(x=Date, y=Period_mean, group=1)) +
  ggtitle("Period Usage") +
  geom_point(color = "dodgerblue3", alpha = 0.7) + 
  geom_smooth(method = "loess", span = 0.50 )+ 
  plot_aes +
  labs(x = "Year", y = '% of Total Words') +
  theme(axis.text.x=element_text(angle=45, hjust=1), 
        plot.title.position = 'plot', 
        plot.title = element_text(hjust = 0.5, face = "bold", size = 16)) +
  theme(plot.title.position = 'plot', 
        plot.title = element_text(hjust = 0.5, face = "bold", size = 16)) + 
  theme(axis.text=element_text(size=16),
        axis.title=element_text(size=20,face="bold"))+
  theme(plot.title.position = 'plot', 
        plot.title = element_text(hjust = 0.5, face = "bold", size = 20)) +
  theme(axis.text=element_text(size = 14),
        axis.title=element_text(size = 20,face="bold")) +
  annotate(geom="text",x=1860,
             y=7,label="
             estimate = 0.0128
             p-value < .001
           
           ", size = 3.5)

#words per sentence
wps_smooth_tidy <- ggplot(data=tidy_df, aes(x=Date, y=WPS_mean, group=1)) +
  ggtitle("Words per Sentence") +
  geom_point(color = "dodgerblue3", alpha = 0.7) + 
  geom_smooth(method = "loess", span = 0.70 )+ 
  plot_aes +
  labs(x = "Year", y = '# of Words') +
  theme(axis.text.x=element_text(angle=45, hjust=1), 
        plot.title.position = 'plot', 
        plot.title = element_text(hjust = 0.5, face = "bold", size = 16)) +
  theme(plot.title.position = 'plot', 
        plot.title = element_text(hjust = 0.5, face = "bold", size = 16)) + 
  theme(axis.text=element_text(size=16),
        axis.title=element_text(size=20,face="bold"))+
  theme(plot.title.position = 'plot', 
        plot.title = element_text(hjust = 0.5, face = "bold", size = 20)) +
  theme(axis.text=element_text(size = 14),
        axis.title=element_text(size = 20,face="bold")) +
  annotate(geom="text",x=1860,
             y=65,label="
             estimate = -0.014
             p-value < .001
           
           ", size = 3.5)


tidy_smooth_graphs <- ggpubr::ggarrange(Analytic_smooth_tidy,Bw_smooth_tidy,
                                  period_smooth_tidy,wps_smooth_tidy,
                                   ncol=2, nrow=2, common.legend = TRUE, legend = "bottom")
annotate_figure(tidy_smooth_graphs,
                top = text_grob("Smooth Formality Graphs (grouped by year)",  color = "black", face = "bold", size = 20),
                bottom = text_grob(
                "Note. Horizontal shading represents Standard Error.
                Estimates are from centered analyses (centered on means from 1857)"
                                   , color = "Black",
                                   hjust = 1.05, x = 1, face = "italic", size = 16))

Make our rough plots (means per year)

Plotting the data by year (one data point per year).

Analytic <- ggplot(data=tidy_df, aes(x=Date, y=Analytic_mean, group=1)) +
   geom_line(colour = "dodgerblue3") +
   geom_ribbon(aes(ymin=Analytic_mean-Analytic_std.error, ymax=Analytic_mean+Analytic_std.error), alpha=0.2) +
   ggtitle("Analytic Thinking") +
   plot_aes + 
   labs(x = "Year", y = 'Standardized score') + 
   theme(axis.text.x=element_text(angle=45, hjust=1), 
         plot.title.position = 'plot', 
         plot.title = element_text(hjust = 0.5, face = "bold", size = 16)) +
   theme(plot.title.position = 'plot', 
         plot.title = element_text(hjust = 0.5, face = "bold", size = 16)) + 
   theme(axis.text=element_text(size=16),
         axis.title=element_text(size=20,face="bold"))+
   theme(plot.title.position = 'plot', 
         plot.title = element_text(hjust = 0.5, face = "bold", size = 20)) +
   theme(axis.text=element_text(size = 14),
         axis.title=element_text(size = 20,face="bold")) 

#WPS 
WPS <- ggplot(data=tidy_df, aes(x=Date, y=WPS_mean, group=1)) +
  geom_line(colour = "dodgerblue3") +
  geom_ribbon(aes(ymin=WPS_mean-WPS_std.error, ymax=WPS_mean+WPS_std.error), alpha=0.2) +
  ggtitle("WPS") +
  plot_aes +
  labs(x = "Year", y = '# of Words') + 
  theme(axis.text.x=element_text(angle=45, hjust=1), 
        plot.title.position = 'plot', 
        plot.title = element_text(hjust = 0.5, face = "bold", size = 16)) +
  theme(plot.title.position = 'plot', 
        plot.title = element_text(hjust = 0.5, face = "bold", size = 16)) + 
  theme(axis.text=element_text(size=16),
        axis.title=element_text(size=20,face="bold"))+
  theme(plot.title.position = 'plot', 
        plot.title = element_text(hjust = 0.5, face = "bold", size = 20)) +
  theme(axis.text=element_text(size = 14),
        axis.title=element_text(size = 20,face="bold")) 

#BigWords 
BigWords <- ggplot(data=tidy_df, aes(x=Date, y=BigWords_mean, group=1)) +
  geom_line(colour = "dodgerblue3") +
  geom_ribbon(aes(ymin=BigWords_mean-BigWords_std.error, ymax=BigWords_mean+BigWords_std.error), alpha=0.2) +
  ggtitle("Big Words N > 6") +
  plot_aes +
  labs(x = "Year", y = '% of Total Words') + 
  theme(axis.text.x=element_text(angle=45, hjust=1), 
        plot.title.position = 'plot', 
        plot.title = element_text(hjust = 0.5, face = "bold", size = 16)) +
  theme(plot.title.position = 'plot', 
        plot.title = element_text(hjust = 0.5, face = "bold", size = 16)) + 
  theme(axis.text=element_text(size=16),
        axis.title=element_text(size=20,face="bold"))+
  theme(plot.title.position = 'plot', 
        plot.title = element_text(hjust = 0.5, face = "bold", size = 20)) +
  theme(axis.text=element_text(size = 14),
        axis.title=element_text(size = 20,face="bold")) 


#period frequency 
Period <- ggplot(data=tidy_df, aes(x=Date, y=Period_mean, group=1)) +
  geom_line(colour = "dodgerblue3") +
  geom_ribbon(aes(ymin=Period_mean-Period_std.error, ymax=Period_mean+Period_std.error), alpha=0.2) +
  ggtitle("Period-usage") +
  plot_aes +
  labs(x = "Year", y = '% of Total Words') +
  theme(axis.text.x=element_text(angle=45, hjust=1), 
        plot.title.position = 'plot', 
        plot.title = element_text(hjust = 0.5, face = "bold", size = 16)) +
  theme(plot.title.position = 'plot', 
        plot.title = element_text(hjust = 0.5, face = "bold", size = 16)) + 
  theme(axis.text=element_text(size=16),
        axis.title=element_text(size=20,face="bold"))+
  theme(plot.title.position = 'plot', 
        plot.title = element_text(hjust = 0.5, face = "bold", size = 20)) +
  theme(axis.text=element_text(size = 14),
        axis.title=element_text(size = 20,face="bold")) 

#raw graphs
raw_graphs <- ggpubr::ggarrange(Analytic,BigWords,Period,WPS,ncol=2, nrow=2, common.legend = TRUE, legend = "bottom")
annotate_figure(raw_graphs,
                top = text_grob("Raw Formality Graphs (grouped by year)",  color = "black", face = "bold", size = 20),
                bottom = text_grob("Note. Horizontal shading represents Standard Error. 
                                   Graphs are of books in the collection"
                                   , color = "Black",
                                   hjust = 1.05, x = 1, face = "italic", size = 16))

Regression Models

Models presented in order: Raw data, aggregated by year, centered on 1857

Analytic Thinking

#Raw Data
AT_RAW <- lm(Analytic ~ Date, data = df)

#Tidy Data
AT_TIDY <- lm(Analytic_mean ~ Date, data = tidy_df)

#centered
AT_centered <- lm(Analytic_centered ~ Date, data = tidy_df)


table_model(AT_RAW)
term estimate SE t p
(Intercept) -114.2 12.7358 -8.965 0
Date 0.1 0.0065 15.333 0 
table_model(AT_TIDY)
term estimate SE t p
(Intercept) -73.5447 21.2185 -3.466 7e-04
Date 0.0782 0.0109 7.152 0e+00 
table_model(AT_centered)
term estimate SE t p
(Intercept) -159.4847 21.2185 -7.516 0
Date 0.0782 0.0109 7.152 0

Big Words (words with a letter count > 6)

BW_Raw <- lm(BigWords ~ Date, data = df)
BW_Tidy <- lm(BigWords_mean ~ Date, data = tidy_df)
BW_centered <- lm(BigWords_centered ~ Date, data = tidy_df)

table_model(BW_Raw)
term estimate SE t p
(Intercept) -114.0403 4.2057 -27.12 0
Date 0.0701 0.0022 32.55 0 
table_model(BW_Tidy)
term estimate SE t p
(Intercept) -62.9412 9.753 -6.454 0
Date 0.0431 0.005 8.569 0 
table_model(BW_centered)
term estimate SE t p
(Intercept) -88.6212 9.753 -9.087 0
Date 0.0431 0.005 8.569 0

Periods

#Periods
Period_Raw <- lm(Period ~ Date, data = df)
Period_Tidy <- lm(Period_mean ~ Date, data = tidy_df)
Period_centered <- lm(Period_centered ~ Date, data = tidy_df)
table_model(Period_Raw)
term estimate SE t p
(Intercept) -34.4258 1.5638 -22.01 0
Date 0.0204 0.0008 25.49 0 
table_model(Period_Tidy)
term estimate SE t p
(Intercept) -20.0945 3.3420 -6.013 0
Date 0.0128 0.0017 7.416 0 
table_model(Period_centered)
term estimate SE t p
(Intercept) -24.6835 3.3420 -7.386 0
Date 0.0128 0.0017 7.416 0

Words per Sentence

#WPS
WPS_Raw <- lm(WPS ~ Date, data = df)
WPS_Tidy <- lm(WPS_mean ~ Date, data = tidy_df)
WPS_centered <- lm(WPS_centered ~ Date, data = tidy_df)

table_model(WPS_Raw)
term estimate SE t p
(Intercept) 125.9095 14.2534 8.834 0
Date -0.0497 0.0073 -6.805 0 
table_model(WPS_Tidy)
term estimate SE t p
(Intercept) 58.434 28.4925 2.0508 0.0419
Date -0.014 0.0147 -0.9535 0.3417 
table_model(WPS_centered)
term estimate SE t p
(Intercept) 21.294 28.4925 0.7473 0.4559
Date -0.014 0.0147 -0.9535 0.3417